Earth Observation commercial data sales have increased a 550% in the last decade. The field is considered a key element in the European Research Road Map and an opportunity market for the next years. The forecast for this decade is $4 billion in commercial data sales at the end of 2019. This makes EO a major field of new business opportunities and work.

Although EO is an established key area for innovation, the access to the information obtained from satellites follows traditional and expensive paths to cover on-demand services for different potential customers: conventional data centres and conventional distribution of services. This presents several drawbacks:

  • The cost of acquiring recent images of the Earth is very high. This is a limitation for small and medium companies to develop new solutions.
  • Clients cannot access the information they need directly nor quickly, because this has to be processed and ad-hoc distributed.
  • The service is not flexible, so does not adapt to sudden changes in demand.

Cloud computing is presented as a possible solution to improve common services and create new market opportunities because it is elastic, scalable and it works on demand through virtualisation of resources [1][2].

Satellite and Earth Observation applications are then clear use cases for deployment on the cloud for the following reasons:

  • The global nature of EO data, with ground stations and users geographically located all over the world, means that it makes sense to deploy a worldwide infrastructure connecting all the stakeholders. Ground stations, ground control centres and data processing centres would be able to take advantage of a rapid, agile, resilient and secure interconnected computer system for sharing the bulk of EO data. Final users would also take advantage of having data access points as close as possible to them to minimise the delay.
  • The massive size of EO data generated by today´s sensors, in the order of daily Terabytes, means that it needs cost-effective procurement of the computing infrastructure for archiving and processing. Thus, it seems a good idea to pay for computer resources only as you need them, as is the usual cost model of services on the cloud. Note that EO data is received in batches for each receiving ground station on each ‘contact’, when a satellite downloads all perceived data since the previous communication. Then the processing, archive and dissemination process of that received data is triggered and executed on the provided infrastructure.
  • A very large amount of resources are needed to optimally process and distribute EO data to the global user community. Data processing would be greatly enhanced by using a massive number of computing nodes working in parallel with techniques related to High Performance Computing (HPC) and High-Throughput Computing (HTC) techniques for, respectively, generating results as fast as possible, and for processing as many jobs as possible in a given time.

However the implementation of these systems with the current cloud computing technology still presents some technical limitations:

  • The virtual machine images (VMIs) are not optimised, being highly oversized. This directly impacts in the costs of using the infrastructure and in the dynamic resources provisioning
  • The deployment of Virtual Machines (VM) in cloud has long duration, normally between 10 and 20 minutes which directly affects the flexibility and dynamic scalability of the systems

Within the ENTICE project framework Deimos’ research focuses in the development of Future Internet technologies in order to improve Earth Observation (EO) services and to highly reduce the costs associated with on-premises deployment. Within the ENTICE H2020 project, Deimos intends to implement a flexible, cost-effective Payload Data Ground Segment (PDGS) in a cloud computing infrastructure that will shorten the latency of the system near the data sources and the final users through an automated process to reduce 60% the VMI size, 30% VMI delivery time, 25% deployment time, 25% VM cost and 25% VM storage to cover the demand for services with highly variable demands [3].

To find out more about ENTICE and EO visit our (use case page)

References

[1] J. Becedas, R. Pérez and G. González, “Testing and validation of cloud infrastructures for Earth observation services with satellite constellations”, International Journal of Remote Sensing,  vol. 36, nº 19-20, pp. 5289-5307, 2015.

[2] J. Becedas, R. Pérez, G. González, J. Álvarez, F. García, F. Maldonado, A. Sucari and J. García, “Evaluation of Future Internet Technologies for Processing and Distribution of Satellite Imagery”, The 36th International Symposium on Remote Sensing of Environment, Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., vol. XL-7/W3, pp. 605-611, doi:10.5194/isprsarchives-XL-7-W3-605-2015, 2015.

[3] J. J. Ramos and J. Becedas, “Deimos’ gs4EO over ENTICE: A cost-effective cloud-based solution to deploy and operate flexible big EO data systems with optimized performance”, Procedings of the 2016 conference on Big Data from Space (BiDS’16), pp. 107-110, Santa Cruz de Tenerife, Spain, 2016.